Reproduce “Latent Heuristic Search: Continuous Optimization for Automated Algorithm Design” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Cheikh Ahmed, Mahdi Mostajabdaveh, Zirui Zhou
LHS searches for traveling-salesperson heuristics by learning continuous program representations and using a ranking surrogate’s gradients to guide candidate generation. A frozen large language model decodes candidates into code, which execution tests. On 100 held-out size-200 instances, LHS reports a mean objective of 12.49 ± 0.20, versus 12.58 ± 0.36 for ReEvo. This edge was not tested for significance.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.